Digital Marketplaces Are Entering the Agentic Era – and the Rules of Commerce Are Being Rewritten
When Amazon told investors on its third-quarter earnings call in late October 2025 that a single AI shopping assistant was on track to generate more than $10 billion in incremental annualized sales, the number landed as more than a corporate milestone. It was a signal. The company disclosed that roughly 250 million shoppers had used the assistant, then called Rufus, during the year, that monthly active users had grown 140% year over year, and that customers who engaged with it were 60% more likely to complete a purchase. Amazon CEO Andy Jassy put it plainly on the call – “Rufus is expected to generate over $10 billion in annual incremental sales for us.” For anyone still treating conversational AI in commerce as a novelty, that framing marked the end of the debate.
The digital marketplace industry has spent two decades optimizing the same fundamental loop, a human browses, filters, compares, and clicks. That loop is now being dismantled and rebuilt. Within a remarkably compressed window in 2025, the largest platforms, payment networks, and cloud providers in the world moved—almost in unison—to prepare their infrastructure for a future in which software agents, not just people, discover products and complete transactions. The question executives should be asking is no longer whether this shift is real. It is how fast their own commerce architecture can adapt to it.
Why the Industry Has Reached an Inflection Point
Inflection points in commerce usually arrive quietly, then all at once. The current one crystallized in the autumn of 2025, when OpenAI and Stripe jointly introduced Instant Checkout in ChatGPT alongside a new open standard called the Agentic Commerce Protocol. The mechanics matter. A ChatGPT user in the United States can now ask for a recommendation, and when ready to buy, complete the purchase inline through a Stripe-powered checkout—initially with U.S.-based Etsy sellers, with more than a million Shopify merchants including Glossier, Vuori, Spanx, and SKIMS to follow.
Stripe’s own description of the change is unusually candid about what it means for existing systems. “Traditional ecommerce was built for humans—businesses controlled the interface and payments while shoppers browsed and clicked,” the company noted, adding that in AI-led commerce, agents now sit between businesses and consumers, so “everything from payments and checkout to fraud checks must be re-architected.” That single sentence captures why this is an inflection point rather than an incremental upgrade. The interface layer that marketplaces spent years perfecting is being disintermediated by a new class of buyer.
This is not confined to one platform. Google introduced its own Agent Payments Protocol in September 2025, and Salesforce announced in October that it would support the Agentic Commerce Protocol with Stripe and OpenAI. When the leading conversational AI company, the dominant search company, the largest enterprise CRM, and the world’s most influential payments infrastructure provider all move toward interoperable agentic commerce standards in the same quarter, that is the market telling enterprises where the puck is heading.
Technologies Reshaping Digital Marketplaces
The technical heart of this transition is a rethinking of how trust and payment credentials travel through a transaction. In the Stripe and OpenAI model, once a buyer confirms a purchase, Stripe issues what it calls a Shared Payment Token—a new payment primitive that lets an application like ChatGPT initiate a payment without exposing the buyer’s underlying card credentials, scoped narrowly to a specific merchant and cart total. That design solves the problem that has quietly blocked autonomous commerce for years – how do you let a machine spend money on your behalf without handing it the keys to your account?
The card networks arrived at the same problem from the other direction. Mastercard unveiled Agent Pay in April 2025, built on tokenization technology it describes as Agentic Tokens, extending the same infrastructure that has long secured digital payments to a world where AI agents act as buyers. Visa moved in parallel with its Intelligent Commerce initiative, positioning itself to let AI agents transact securely across the payment ecosystem. The strategic logic is straightforward. If agent-led transactions become the norm, whichever network defines the trust layer for those transactions retains its position at the center of digital commerce. Neither incumbent intends to be disintermediated.
For enterprise buyers and marketplace operators, the practical implication is that the technology stack is fragmenting into competing but increasingly interoperable protocols. The winners will not be those who bet on a single standard, but those who build commerce architecture flexible enough to expose products, pricing, and checkout across whatever agent surfaces their customers happen to use. As Stripe noted, with many AI agents emerging, it is simply not realistic for a business to maintain a bespoke integration with each one.
AI’s Expanding Role Across Marketplace Operations
Agentic checkout is the most dramatic development, but it sits atop a broader wave of AI adoption that is already reshaping how marketplaces operate day to day. Amazon’s disclosure is instructive precisely because it quantifies something most companies still discuss in the abstract. The retailer trained its assistant on its entire product catalog, customer reviews, community questions, and information from across the web, then measured the downstream commercial impact through a seven-day rolling attribution model. The strategic intent is telling – Amazon is using AI to keep shoppers doing their product research inside its own ecosystem rather than defecting to a search engine or an external chatbot where competitors lurk.
Shopify’s numbers point in the same direction from the merchant side of the equation. The company has reported that orders arriving through AI-driven search have climbed roughly thirteenfold since the start of 2025, and it launched an Agentic Plan that lets brands—even those hosted on other platforms—sync products into the Shopify Catalog and sell through AI channels. Two of the most influential companies in commerce are effectively telling the market the same thing from opposite ends of the value chain, AI is no longer a support function bolted onto the marketplace. It is becoming the discovery surface itself.
That reframing has consequences for merchandising, advertising, and customer service alike. Product discovery is migrating from keyword search toward conversational intent, which means the metadata, structured pricing, and machine-readable catalogs a seller maintains now determine whether an agent can even find and recommend a product. The marketplaces that thrive will be those that treat their catalog as an API surface for machines, not merely a webpage for humans.
Digital Trust, Security, and Regulatory Compliance
As commerce becomes faster and more automated, the burden of trust rises rather than falls. Regulators recognized this well before the agentic wave crested. Under the European Union’s Digital Services Act, online marketplaces are required to verify the identity of the traders selling on their platforms through so-called Know Your Business Customer obligations, and to design their systems so that sellers can readily comply with legal requirements—a “compliance by design” mandate embedded in the regulation. The intent is to create a safer, more transparent online environment in which counterfeit and unsafe goods have fewer places to hide.
These requirements were written for a world of human buyers and identifiable sellers. Agentic commerce complicates both sides of that relationship. When an autonomous agent initiates a purchase, questions of consent, liability, and fraud attribution become more intricate. This is precisely why the payment primitives introduced in 2025 place so much emphasis on scoping tokens to a specific merchant and a specific cart total, and on preserving fraud protection even when the payment is processed elsewhere. Marketplace security, identity verification, and seller vetting are shifting from back-office hygiene to core competitive infrastructure. Enterprises that treat compliance as a checkbox rather than an architectural principle will find themselves exposed as both regulation and automation intensify.
Cross-Border Commerce Under a New Trade Regime
The economics of cross-border commerce also changed abruptly in 2025. The United States eliminated the long-standing de minimis exemption that had allowed shipments valued under $800 to enter the country free of duties and taxes, closing the treatment for goods from China and Hong Kong in the spring and ending it more broadly on August 29, 2025. That exemption was the mechanism that underpinned the direct-from-factory model popularized by fast-growing marketplaces, and its removal added substantial costs to low-value international parcels, with platforms including Etsy, eBay, and Shein among those affected.
For strategic buyers and investors, this is a structural shift, not a temporary tariff skirmish. The regulatory tailwind that made ultra-low-cost cross-border commerce viable has weakened, and the European Union is moving to abolish its own low-value exemption as well. Marketplaces that built their unit economics on duty-free micro-parcels now face a genuine reckoning, while platforms with diversified fulfillment, domestic inventory, and stronger margins gain relative advantage. Cross-border commerce is not disappearing, but its cost structure is being reset—and enterprises should model that reset into their sourcing and logistics strategies rather than assume the old arithmetic still holds.
Enterprise Strategies and Investment Signals to Watch
Read together, these developments describe a market where value is migrating toward three layers, the AI discovery surface, the trust-and-payment infrastructure that authorizes agent-led transactions, and the catalog architecture that makes products legible to machines. The clearest investment signal of 2025 was the sheer coordination of incumbents. When Amazon quantifies a $10 billion AI-driven sales lift, Shopify reports AI orders up thirteenfold, Stripe and OpenAI ship an open protocol, Google publishes a competing one, and Visa and Mastercard both race to define agentic payment tokens, capital and talent follow that gravity.
For enterprise leaders, the practical priorities are becoming concrete. Structured, machine-readable product data is now a revenue driver, not a technical nicety. Payment and identity architecture must be flexible enough to support tokenized, agent-initiated transactions across multiple protocols. Compliance and seller verification, sharpened by the Digital Services Act, should be engineered in rather than retrofitted. And cross-border strategy needs to be rebuilt around a trade environment in which the de minimis subsidy is gone. Marketplace operators that move early on these fronts will be positioned to capture the demand agents redirect; those that wait risk becoming invisible to the very software now doing the shopping.
What the Next Five Years Could Look Like
The trajectory is not difficult to read even if the pace remains uncertain. Discovery will continue migrating from search boxes to conversation. Payments will increasingly be authorized by tokens scoped to agents rather than typed by humans into forms. Regulation will keep tightening around identity, safety, and accountability as automation removes friction faster than it removes risk. And the cost of cross-border commerce will settle at a higher, more sustainable level than the exemption-fueled era allowed.
What separates the marketplace leaders of the coming decade from the rest will not be who adopted AI first, but who rebuilt their commerce architecture around the reality that the buyer may increasingly be a machine acting on a human’s behalf. The companies quantifying real revenue from that shift today—Amazon, Shopify, Stripe, the card networks—are not making speculative bets. They are describing the operating conditions everyone else in the Digital Marketplace ecosystem will soon share. The strategic task for executives, investors, and founders is to stop treating agentic commerce as a headline and start treating it as the new grammar of the market. Those who learn that language early will define the next phase. Those who don’t may find their catalogs perfectly optimized for a shopper who no longer exists.
